Showing posts with label fractal. Show all posts
Showing posts with label fractal. Show all posts

Friday, May 13, 2011

Discrete connections (part XII)



First a small stop to explain this sudden "gold rush" about finite continued fractions and how they impact upon the subject at stake.

1. As mentioned previously, they are related to the euclidean fractal by its construction. Indeed, consider:
[; the \; (p_{s}-1)^{th} \; block \; of \; \ldots \; of \; the \; (p_{1}-1)^{th} \; block \; of ;]
[; the \; euclidean \; fractal \; defined \; by \; u=0, v=i, w=1+i, z=1 ;]
then its base endpoints' real parts are:
[;\cfrac{1}{p_1+\cfrac{1}{\ddots + \cfrac{1}{p_s}}}} \; and \; \cfrac{1}{p_1+\cfrac{1}{\ddots + \cfrac{1}{p_s + 1}}}};]

2. As you might already know, any positive rational number can be expressed as a continued fraction whose components are, in a top-down representation, the successive quotients obtained through Euclid's gcd algorithm applied to the numerator and denominator. Moreover, there is only one such canonical representation, i.e. if a sub-unitary rational number k/n is equal to a canonical finite continued fraction then we have all the steps of Euclid's gcd algorithm (we have all the quotients) for n and k.

Further more, if we strictly bound a positive sub-unitary rational number k/n to values such as described in 1, then Euclid's gcd algorithm for n and k has at least s steps, for which the quotients are given by the continued fractions defining the boundaries.

Tuesday, February 1, 2011

Discrete connections (part VI)


The main block of the euclidean fractal
For any four complex numbers u,v,w and z shaping a trapezoid let us consider the entity defined by the following elements:
the two base endpoints
[; u \; and \; z;]
the two top endpoints
[;v \; and \; w;]
the top line
[;\delta(v,w);]
the parallelism master line
[;\delta(u,v);]
the top value
[;\beta \in \sigma(v,w) \; such \; that \; \delta(\alpha,\beta) \| \delta(u,v);]
where:
[;\delta(z_1,z_2) \; - \; the \; complex \; line \; defined \; by \; z_1 \; and \; z_2;]
[;\sigma(z_1,z_2) \; - \; the \; open \; complex \; line \; segment \; defined \; by \; z_1 \; and \; z_2;]
[;\alpha \; -\; the \; limit \; of \; the \; top \; value \; base \; sequence ;]
The self similarity block of rank k of the euclidean fractal
For any four complex numbers u,v,w and z shaping a trapezoid let us consider the entity defined by the following elements:
the two base endpoints
[;\tilde{\alpha}_{k+1} \; and \; \tilde{\beta}_{k+2};]
the two top endpoints
[;\tilde{\beta}_{k+1} \; and \; 2\cdot \tilde{\beta}_{k+2} - \tilde{\alpha}_{k+2};]
the top line
[;\delta(\tilde{\beta}_{k+1}, 2 \cdot \tilde{\beta}_{k+2} - \tilde{\alpha}_{k+2});]
the parallelism master line
[;\delta(\tilde{\alpha}_{k+1}, \tilde{\beta}_{k+1});]
the top value
[;\beta^{(k)} = \overline{\sigma}(\tilde{\alpha}_0,\beta) \cap \sigma(\tilde{\beta}_{k+1}, 2 \cdot \tilde{\beta}_{k+2} - \tilde{\alpha}_{k+2});]
where:
[;\beta \; - \; defined \; as \; above;]
[;\overline{\sigma}(z_1,z_2) \; the \; closed \; complex \; line \; segment \; defined \; by \; z_1 \; and \; z_2;]
Moreover, the self similarity block of rank k is by itself an euclidean fractal main block.
Now for its link to the euclidean summation, consider the euclidean fractal main block defined by u=0,v=i,w=1+i and z=1. (suggestive figure pending)
Next: Fibonacci decomposition of n

Saturday, January 29, 2011

Discrete connections (part V)


This part will deal mainly with the constructive elements of a fractal class derived from the conjectured asymptotic nature of the euclidean summation.
First some basic sequences and functions:
r-golden ratio builder
[;g_0^r=r, \; g_{n+1}^r = 1 + \frac{1}{g_n^r};]
upward shifted by r identity
[;u_n^r=n+r;]
upper ratio coefficient
[;\overline{\lambda}:[0,\infty) \rightarrow (0,1], \; \overline{\lambda}(x)=\frac{1}{1+x};]
where r>0 is a real number.
Let us consider some properties of the above sequences and functions:
[;\lim_{n\rightarrow \infty}\overline{\lambda}(g_n^r) = 1-\phi, \sum_n (-1)^n \cdot \prod_{i=0}^{n} \overline{\lambda}(g_i^r) \; convergent;]
[;\lim_{n \rightarrow \infty} \overline{\lambda}(u_n^r) = 0, \prod_n (1 - \overline{\lambda}(u_n^r)) \; convergent \; to \; 0;]
For the upward shifted by r identity, the conclusions are self evident due to:
[;\overline{\lambda}(u_n^r) = \frac{1}{1+n+r};]
and
[;\prod_{i=0}^{n} (1 - \overline{\lambda}(u_n^r)) = \frac{r}{1+n+r};]
For the r-golden ratio builder, the first conclusion's proof is based upon the relative positions of r and the golden ratio.
Indeed, if r is equal to the golden ratio then the r-golden ratio builder is stationary (and equal to the golden ratio).
From the two left cases let's consider the case r is smaller than the golden ratio. With this assumption we get:
[;r^2 -r -1 < 0 \Rightarrow g_1^r - g_0^r = -\frac{r^2-r-1}{r} > 0;] 
By means of mathematical induction one can prove that the subsequences of even index and odd index terms of the r-golden ratio are respectively increasing and decreasing. Since both sequences fall between the two first terms of the sequence results that they are both convergent to respectively g and h. 
We get successively:
[;\left\{ \begin{array}{rcl} g & = &1+\frac{1}{h}\\ h & = &1+\frac{1}{g} \end{array} \right. \Leftrightarrow \left\{ \begin{array}{rcl} g & = &1+\frac{1}{1+\frac{1}{g}} \\ h & = & 1+\frac{1}{g} \end{array} \right. \Leftrightarrow;]
[;\left\{ \begin{array}{rcl}g & = & \frac{2\cdot g +1}{g+1} \\ h & = & 1 + \frac{1}{g}\end{array}\right. \Leftrightarrow \left\{ \begin{array}{rcl} g^2 -g -1 & = & 0\\ h & = & 1 + \frac{1}{g} \end{array} \right. \Leftrightarrow;]
[;\left\{ \begin{array}{rcl}g & = & \varphi \\ h & = & 1 + \frac{1}{g}\end{array}\right. \Leftrightarrow \left\{ \begin{array}{rcl} g & = & \varphi \\ h & = & \varphi \end{array} \right.;]
Therefore the entire sequence is convergent to the golden ratio. Since the golden ratio is not equal to 1 results that:
[;\sum_n (-1)^n \cdot \prod_{i=0}^{n} \overline{\lambda}(g_i^r);]
is convergent.
Having these sequences and functions completely described let us consider the two mechanisms we will use in order to define the previously mentioned class of fractals:
The top value base sequence
For any four complex numbers u,v,w,z shaping a trapezoid and the sequences:
[;(\alpha_n)_{n\ge 0}, \; (\beta_n)_{n\ge 0};]
defined by:
[;\alpha_0=u, \; \beta_0=v, \beta_1=w,\; \alpha_1=z;]
and
[;\left\{ \begin{array}{rcl} \alpha_{n+2} & = & (1 - \overline{\lambda}(g_n^r))\cdot \alpha_{n+1}+ \overline{\lambda}(g_n^r) \cdot \alpha_{n} \\ \beta_{n+2} & = & (1 - \overline{\lambda}(g_n^r))\cdot \alpha_{n+1}+ \overline{\lambda}(g_n^r) \cdot \beta_{n} \end{array} \right. ;]
we have that:
[;(\alpha_n)_{n\ge 0} \rightarrow \alpha;]
The self-similarity base sequence
For any four complex numbers u,v,w,z shaping a trapezoid and the sequences:
[;(\tilde{\alpha}_n)_{n\ge 0}, \; (\tilde{\beta}_n)_{n\ge 0};]
defined by:
[;\tilde{\alpha}_0=u, \; \tilde{\beta}_0=v, \tilde{\beta}_1=w,\; \tilde{\alpha}_1=z;]
and
[;\left\{ \begin{array}{rcl} \tilde{\alpha}_{n+2} & = & (1 - \overline{\lambda}(u_n^r))\cdot \tilde{\alpha}_{n+1}+ \overline{\lambda}(u_n^r) \cdot \tilde{\alpha}_{0} \\ \tilde{\beta}_{n+2} & = & (1 - \overline{\lambda}(u_n^r))\cdot \tilde{\alpha}_{n+1}+ \overline{\lambda}(u_n^r) \cdot \tilde{\beta}_{0} \end{array} \right. ;]
we have that:
[;(\tilde{\alpha}_n)_{n\ge 0} \rightarrow \tilde{\alpha}_0;]
For both sequences r is the scaling ratio of the two parallel edges of the trapezoid i.e.:
[;(u-v)=r\cdot (z-w);]
The construction of the above two sets of sequences is better depicted by the following figures:
and
Next: The euclidean fractal class

Monday, January 24, 2011

Discrete connections (part II)


First, some Scilab-powered graphic suggesting the conjectured fractal nature of the euclidean summation:
The normalized euclidean summation (m=10000 and 0<n<10000)

But science relies upon facts, so let's begin by listing some of the basic properties of the euclidean summation:
Lower boundary
[;\tau_1(n,k) = 0 \Leftrightarrow (k=0)$ or $((k\neq 0)$ and $(k|n));]
Homogeneity
[;\tau_1(p\cdot n, p\cdot k) = p \cdot \tau_1(n,k);] 
Column periodicity
[;m_1\equiv m_2\pmod{n}$ $\Rightarrow$ $\tau_1(m_1, n) = \tau_1(m_2,n);]
Outer row recurrence
[;\frac{n}{2}<k<n \Rightarrow \tau_1(n,k) = n-k + \tau_1(k,n-k);]
Inner row recurrence
[;\frac{n}{2}<k<n \Rightarrow \tau_1(n,k) = n-k + \tau_1(n,n-k);]
Diagonal periodicity
[;\frac{n}{2}<k<n \; and \; 0 \le r < (n-k) \Rightarrow ;]
[;\tau_1(n + p \cdot (n-k) + r, k + p \cdot (n-k) + r) = \tau_1(n+r, k+r) ;]

Next: Boundaries for peak and positional peak

Saturday, January 22, 2011

Fractals and life (part I)

Well, these are a special breed of mathematical beasts, they are ever growing collections of dots and no matter how deep you zoom them in you'll get about the same picture as you got when looking at the original.
As an example, one of the most intuitive ones, we will consider Sierpinski's triangle: let us take a triangle with all sides equal and join the middles of its sides. We thus get four smaller triangles. Lets cut off the one in the center. We now have only three triangles in the corners, and repeat for each one of them the same procedure: join middles and cut the center triangle. If we infinitely repeat the procedure we would obtain something that looks like this:

Where does life use these kind of tools? Frosted window patterns, broccoli, shattered compact discs, Scandinavian fjords and the list goes on. 

References: